IntroUNET: Identifying introgressed alleles via semantic segmentation.

A growing body of evidence suggests that gene flow between closely related species is a widespread phenomenon. Alleles that introgress from one species into a close relative are typically neutral or deleterious, but sometimes confer a significant fitness advantage. Given the potential relevance to speciation and adaptation, numerous methods have therefore been devised to identify regions of the genome that have experienced introgression. Recently, supervised machine learning approaches have been shown to be highly effective for detecting introgression. One especially promising approach is to treat population genetic inference as an image classification problem, and feed an image representation of a population genetic alignment as input to a deep neural network that distinguishes among evolutionary models (i.e. introgression or no introgression). However, if we wish to investigate the full extent and fitness effects of introgression, merely identifying genomic regions in a population genetic alignment that harbor introgressed loci is insufficient-ideally we would be able to infer precisely which individuals have introgressed material and at which positions in the genome. Here we adapt a deep learning algorithm for semantic segmentation, the task of correctly identifying the type of object to which each individual pixel in an image belongs, to the task of identifying introgressed alleles. Our trained neural network is thus able to infer, for each individual in a two-population alignment, which of those individual's alleles were introgressed from the other population. We use simulated data to show that this approach is highly accurate, and that it can be readily extended to identify alleles that are introgressed from an unsampled "ghost" population, performing comparably to a supervised learning method tailored specifically to that task. Finally, we apply this method to data from Drosophila, showing that it is able to accurately recover introgressed haplotypes from real data. This analysis reveals that introgressed alleles are typically confined to lower frequencies within genic regions, suggestive of purifying selection, but are found at much higher frequencies in a region previously shown to be affected by adaptive introgression. Our method's success in recovering introgressed haplotypes in challenging real-world scenarios underscores the utility of deep learning approaches for making richer evolutionary inferences from genomic data.

Tree sequences as a general-purpose tool for population genetic inference.

As population genetics data increases in size new methods have been developed to store genetic information in efficient ways, such as tree sequences. These data structures are computationally and storage efficient, but are not interchangeable with existing data structures used for many population genetic inference methodologies such as the use of convolutional neural networks (CNNs) applied to population genetic alignments. To better utilize these new data structures we propose and implement a graph convolutional network (GCN) to directly learn from tree sequence topology and node data, allowing for the use of neural network applications without an intermediate step of converting tree sequences to population genetic alignment format. We then compare our approach to standard CNN approaches on a set of previously defined benchmarking tasks including recombination rate estimation, positive selection detection, introgression detection, and demographic model parameter inference. We show that tree sequences can be directly learned from using a GCN approach and can be used to perform well on these common population genetics inference tasks with accuracies roughly matching or even exceeding that of a CNN-based method. As tree sequences become more widely used in population genetics research we foresee developments and optimizations of this work to provide a foundation for population genetics inference moving forward.

Facultative parthenogenesis discovered for the first time in Jamaican boa (Chilabothrus subflavus) using novel microsatellite markers.

The mode of reproduction most often seen in snakes is sexual, but studies have noted facultative parthenogenesis in at least six families. Here, we provide evidence for the first observed case of facultative parthenogenesis in a captive Jamaican boa (Chilabothrus subflavus). A 7-year-old female Jamaican boa, isolated since birth, was found to have produced a litter of 15 offspring. To provide molecular DNA evidence of parthenogenesis, 13 new microsatellite loci were isolated in the species. All offspring were found to be homozygous at each locus and only possess alleles found in the dam, implicating that they were born from asexual reproduction. Several developmental abnormalities, including stillbirths and spinal deformities, were noted in the litter which may be explained by their increased level of homozygosity. To preserve genetic diversity in the captive population, research should be conducted to understand the prevalence of this mode of reproduction and to guide future management decisions of this IUCN listed Vulnerable species.

FUT2 Variants Confer Susceptibility to Familial Otitis Media.

Non-secretor status due to homozygosity for the common FUT2 variant c.461G>A (p.Trp154∗) is associated with either risk for autoimmune diseases or protection against viral diarrhea and HIV. We determined the role of FUT2 in otitis media susceptibility by obtaining DNA samples from 609 multi-ethnic families and simplex case subjects with otitis media. Exome and Sanger sequencing, linkage analysis, and Fisher exact and transmission disequilibrium tests (TDT) were performed. The common FUT2 c.604C>T (p.Arg202∗) variant co-segregates with otitis media in a Filipino pedigree (LOD = 4.0). Additionally, a rare variant, c.412C>T (p.Arg138Cys), is associated with recurrent/chronic otitis media in European-American children (p = 1.2 × 10-5) and US trios (TDT p = 0.01). The c.461G>A (p.Trp154∗) variant was also over-transmitted in US trios (TDT p = 0.01) and was associated with shifts in middle ear microbiota composition (PERMANOVA p < 10-7) and increased biodiversity. When all missense and nonsense variants identified in multi-ethnic US trios with CADD > 20 were combined, FUT2 variants were over-transmitted in trios (TDT p = 0.001). Fut2 is transiently upregulated in mouse middle ear after inoculation with non-typeable Haemophilus influenzae. Four FUT2 variants-namely p.Ala104Val, p.Arg138Cys, p.Trp154∗, and p.Arg202∗-reduced A antigen in mutant-transfected COS-7 cells, while the nonsense variants also reduced FUT2 protein levels. Common and rare FUT2 variants confer susceptibility to otitis media, likely by modifying the middle ear microbiome through regulation of A antigen levels in epithelial cells. Our families demonstrate marked intra-familial genetic heterogeneity, suggesting that multiple combinations of common and rare variants plus environmental factors influence the individual otitis media phenotype as a complex trait.

Collective Pulsing in Xeniid Corals: Part I-Using Computer Vision and Information Theory to Search for Coordination.

Xeniid corals (Cnidaria: Alcyonacea), a family of soft corals, include species displaying a characteristic pulsing behavior. This behavior has been shown to increase oxygen diffusion away from the coral tissue, resulting in higher photosynthetic rates from mutualistic symbionts. Maintaining such a pulsing behavior comes at a high energetic cost, and it has been proposed that coordinating the pulse of individual polyps within a colony might enhance the efficiency of fluid transport. In this paper, we test whether patterns of collective pulsing emerge in coral colonies and investigate possible interactions between polyps within a colony. We video recorded different colonies of Heteroxenia sp. in a laboratory environment. Our methodology is based on the systematic integration of a computer vision algorithm (ISOMAP) and an information-theoretic approach (transfer entropy), offering a vantage point to assess coordination in collective pulsing. Perhaps surprisingly, we did not detect any form of collective pulsing behavior in the colonies. Using artificial data sets, however, we do demonstrate that our methodology is capable of detecting even weak information transfer. The lack of a coordination is consistent with previous work on many cnidarians where coordination between actively pulsing polyps and medusa has not been observed. In our companion paper, we show that there is no fluid dynamic benefit of coordinated pulsing, supporting this result. The lack of coordination coupled with no obvious fluid dynamic benefit to grouping suggests that there may be non-fluid mechanical advantages to forming colonies, such as predator avoidance and defense.

A2ML1 and otitis media: novel variants, differential expression, and relevant pathways.

A genetic basis for otitis media is established, however, the role of rare variants in disease etiology is largely unknown. Previously a duplication variant within A2ML1 was identified as a significant risk factor for otitis media in an indigenous Filipino population and in US children. In this report exome and Sanger sequencing was performed using DNA samples from the indigenous Filipino population, Filipino cochlear implantees, US probands, Finnish, and Pakistani families with otitis media. Sixteen novel, damaging A2ML1 variants identified in otitis media patients were rare or low-frequency in population-matched controls. In the indigenous population, both gingivitis and A2ML1 variants including the known duplication variant and the novel splice variant c.4061 + 1 G>C were independently associated with otitis media. Sequencing of salivary RNA samples from indigenous Filipinos demonstrated lower A2ML1 expression according to the carriage of A2ML1 variants. Sequencing of additional salivary RNA samples from US patients with otitis media revealed differentially expressed genes that are highly correlated with A2ML1 expression levels. In particular, RND3 is upregulated in both A2ML1 variant carriers and high-A2ML1 expressors. These findings support a role for A2ML1 in keratinocyte differentiation within the middle ear as part of otitis media pathology and the potential application of ROCK inhibition in otitis media.

A novel mechanism of mixing by pulsing corals.

The dynamic pulsation of xeniid corals is one of the most fascinating phenomena observed in coral reefs. We quantify for the first time the flow near the tentacles of these soft corals, the active pulsations of which are thought to enhance their symbionts' photosynthetic rates by up to an order of magnitude. These polyps are approximately 1 cm in diameter and pulse at frequencies between approximately 0.5 and 1 Hz. As a result, the frequency-based Reynolds number calculated using the tentacle length and pulse frequency is on the order of 10 and rapidly decays as with distance from the polyp. This introduces the question of how these corals minimize the reversibility of the flow and bring in new volumes of fluid during each pulse. We estimate the Péclet number of the bulk flow generated by the coral as being on the order of 100-1000 whereas the flow between the bristles of the tentacles is on the order of 10. This illustrates the importance of advective transport in removing oxygen waste. Flow measurements using particle image velocimetry reveal that the individual polyps generate a jet of water with positive vertical velocities that do not go below 0.1 cm s-1 and with average volumetric flow rates of approximately 0.71 cm3 s-1 Our results show that there is nearly continual flow in the radial direction towards the polyp with only approximately 3.3% back flow. 3D numerical simulations uncover a region of slow mixing between the tentacles during expansion. We estimate that the average flow that moves through the bristles of the tentacles is approximately 0.03 cm s-1 The combination of nearly continual flow towards the polyp, slow mixing between the bristles, and the subsequent ejection of this fluid volume into an upward jet ensures the polyp continually samples new water with sufficient time for exchange to occur.

Three-dimensional trajectories and network analyses of group behaviour within chimney swift flocks during approaches to the roost.

Chimney swifts (Chaetura pelagica) are highly manoeuvrable birds notable for roosting overnight in chimneys, in groups of hundreds or thousands of birds, before and during their autumn migration. At dusk, birds gather in large numbers from surrounding areas near a roost site. The whole flock then employs an orderly, but dynamic, circling approach pattern before rapidly entering a small aperture en masse We recorded the three-dimensional trajectories of ≈1 800 individual birds during a 30 min period encompassing flock formation, circling, and landing, and used these trajectories to test several hypotheses relating to flock or group behaviour. Specifically, we investigated whether the swifts use local interaction rules based on topological distance (e.g. the n nearest neighbours, regardless of their distance) rather than physical distance (e.g. neighbours within x m, regardless of number) to guide interactions, whether the chimney entry zone is more or less cooperative than the surrounding flock, and whether the characteristic subgroup size is constant or varies with flock density. We found that the swift flock is structured around local rules based on physical distance, that subgroup size increases with density, and that there exist regions of the flock that are less cooperative than others, in particular the chimney entry zone.

IntroUNET: identifying introgressed alleles via semantic segmentation.

A growing body of evidence suggests that gene flow between closely related species is a widespread phenomenon. Alleles that introgress from one species into a close relative are typically neutral or deleterious, but sometimes confer a significant fitness advantage. Given the potential relevance to speciation and adaptation, numerous methods have therefore been devised to identify regions of the genome that have experienced introgression. Recently, supervised machine learning approaches have been shown to be highly effective for detecting introgression. One especially promising approach is to treat population genetic inference as an image classification problem, and feed an image representation of a population genetic alignment as input to a deep neural network that distinguishes among evolutionary models (i.e. introgression or no introgression). However, if we wish to investigate the full extent and fitness effects of introgression, merely identifying genomic regions in a population genetic alignment that harbor introgressed loci is insufficient-ideally we would be able to infer precisely which individuals have introgressed material and at which positions in the genome. Here we adapt a deep learning algorithm for semantic segmentation, the task of correctly identifying the type of object to which each individual pixel in an image belongs, to the task of identifying introgressed alleles. Our trained neural network is thus able to infer, for each individual in a two-population alignment, which of those individual's alleles were introgressed from the other population. We use simulated data to show that this approach is highly accurate, and that it can be readily extended to identify alleles that are introgressed from an unsampled "ghost" population, performing comparably to a supervised learning method tailored specifically to that task. Finally, we apply this method to data from Drosophila, showing that it is able to accurately recover introgressed haplotypes from real data. This analysis reveals that introgressed alleles are typically confined to lower frequencies within genic regions, suggestive of purifying selection, but are found at much higher frequencies in a region previously shown to be affected by adaptive introgression. Our method's success in recovering introgressed haplotypes in challenging real-world scenarios underscores the utility of deep learning approaches for making richer evolutionary inferences from genomic data.

Tree sequences as a general-purpose tool for population genetic inference.

As population genetics data increases in size new methods have been developed to store genetic information in efficient ways, such as tree sequences. These data structures are computationally and storage efficient, but are not interchangeable with existing data structures used for many population genetic inference methodologies such as the use of convolutional neural networks (CNNs) applied to population genetic alignments. To better utilize these new data structures we propose and implement a graph convolutional network (GCN) to directly learn from tree sequence topology and node data, allowing for the use of neural network applications without an intermediate step of converting tree sequences to population genetic alignment format. We then compare our approach to standard CNN approaches on a set of previously defined benchmarking tasks including recombination rate estimation, positive selection detection, introgression detection, and demographic model parameter inference. We show that tree sequences can be directly learned from using a GCN approach and can be used to perform well on these common population genetics inference tasks with accuracies roughly matching or even exceeding that of a CNN-based method. As tree sequences become more widely used in population genetics research we foresee developments and optimizations of this work to provide a foundation for population genetics inference moving forward.

Mechanical and biochemical feedback combine to generate complex contractile oscillations in cytokinesis.

The actomyosin cortex is an active material that generates force to drive shape changes via cytoskeletal remodeling. Cytokinesis is the essential cell division event during which a cortical actomyosin ring closes to separate two daughter cells. Our active gel theory predicted that actomyosin systems controlled by a biochemical oscillator and experiencing mechanical strain would exhibit complex spatiotemporal behavior. To test whether active materials in vivo exhibit spatiotemporally complex kinetics, we imaged the C. elegans embryo with unprecedented temporal resolution and discovered that sections of the cytokinetic cortex undergo periodic phases of acceleration and deceleration. Contractile oscillations exhibited a range of periodicities, including those much longer periods than the timescale of RhoA pulses, which was shorter in cytokinesis than in any other biological context. Modifying mechanical feedback in vivo or in silico revealed that the period of contractile oscillation is prolonged as a function of the intensity of mechanical feedback. Fast local ring ingression occurs where speed oscillations have long periods, likely due to increased local stresses and, therefore, mechanical feedback. Fast ingression also occurs where material turnover is high, in vivo and in silico. We propose that downstream of initiation by pulsed RhoA activity, mechanical feedback, including but not limited to material advection, extends the timescale of contractility beyond that of biochemical input and, therefore, makes it robust to fluctuations in activation. Circumferential propagation of contractility likely allows for sustained contractility despite cytoskeletal remodeling necessary to recover from compaction. Thus, like biochemical feedback, mechanical feedback affords active materials responsiveness and robustness.

This population does not exist: learning the distribution of evolutionary histories with generative adversarial networks.

Numerous studies over the last decade have demonstrated the utility of machine learning methods when applied to population genetic tasks. More recent studies show the potential of deep-learning methods in particular, which allow researchers to approach problems without making prior assumptions about how the data should be summarized or manipulated, instead learning their own internal representation of the data in an attempt to maximize inferential accuracy. One type of deep neural network, called Generative Adversarial Networks (GANs), can even be used to generate new data, and this approach has been used to create individual artificial human genomes free from privacy concerns. In this study, we further explore the application of GANs in population genetics by designing and training a network to learn the statistical distribution of population genetic alignments (i.e. data sets consisting of sequences from an entire population sample) under several diverse evolutionary histories-the first GAN capable of performing this task. After testing multiple different neural network architectures, we report the results of a fully differentiable Deep-Convolutional Wasserstein GAN with gradient penalty that is capable of generating artificial examples of population genetic alignments that successfully mimic key aspects of the training data, including the site-frequency spectrum, differentiation between populations, and patterns of linkage disequilibrium. We demonstrate consistent training success across various evolutionary models, including models of panmictic and subdivided populations, populations at equilibrium and experiencing changes in size, and populations experiencing either no selection or positive selection of various strengths, all without the need for extensive hyperparameter tuning. Overall, our findings highlight the ability of GANs to learn and mimic population genetic data and suggest future areas where this work can be applied in population genetics research that we discuss herein.

Mechanical positive feedback and biochemical negative feedback combine to generate complex contractile oscillations in cytokinesis.

Contractile force generation by the cortical actomyosin cytoskeleton is essential for a multitude of biological processes. The actomyosin cortex behaves as an active material that drives local and large-scale shape changes via cytoskeletal remodeling in response to biochemical cues and feedback loops. Cytokinesis is the essential cell division event during which a cortical actomyosin ring generates contractile force to change cell shape and separate two daughter cells. Our recent work with active gel theory predicts that actomyosin systems under the control of a biochemical oscillator and experiencing mechanical strain will exhibit complex spatiotemporal behavior, but cytokinetic contractility was thought to be kinetically simple. To test whether active materials in vivo exhibit spatiotemporally complex kinetics, we used 4-dimensional imaging with unprecedented temporal resolution and discovered sections of the cytokinetic cortex undergo periodic phases of acceleration and deceleration. Quantification of ingression speed oscillations revealed wide ranges of oscillation period and amplitude. In the cytokinetic ring, activity of the master regulator RhoA pulsed with a timescale of approximately 20 seconds, shorter than that reported for any other biological context. Contractility oscillated with 20-second periodicity and with much longer periods. A combination of in vivo and in silico approaches to modify mechanical feedback revealed that the period of contractile oscillation is prolonged as a function of the intensity of mechanical feedback. Effective local ring ingression is characterized by slower speed oscillations, likely due to increased local stresses and therefore mechanical feedback. Fast ingression also occurs where material turnover is high, in vivo and in silico . We propose that downstream of initiation by pulsed RhoA activity, mechanical positive feedback, including but not limited to material advection, extends the timescale of contractility beyond that of biochemical input and therefore makes it robust to fluctuations in activation. Circumferential propagation of contractility likely allows sustained contractility despite cytoskeletal remodeling necessary to recover from compaction. Our work demonstrates that while biochemical feedback loops afford systems responsiveness and robustness, mechanical feedback must also be considered to describe and understand the behaviors of active materials in vivo .

Novel Variants in Hearing Loss Genes and Associations With Audiometric Thresholds in a Multi-ethnic Cohort of US Patients With Cochlear Implants.

OBJECTIVES: To investigate novel variants in hearing loss genes and clinical factors affecting audiometric outcomes of cochlear implant (CI) patients. BACKGROUND: Approximately 50% of hearing loss has a genetic etiology, with certain genetic variants more prevalent in specific ethnic groups. Different variants and some clinical variables including inner ear malformations result in different prognoses or clinical outcomes after CI. METHODS: Medical and genetic testing records of pediatric CI patients were reviewed for clinical variables. Minor allele frequencies of variants were obtained from Genome Aggregation Database (gnomAD) and variants were classified for pathogenicity. Standard statistical testing was done using Fisher's exact, Wilcoxon, and Spearman correlation tests. RESULTS: Eighteen CI patients with genetic test results had pathogenic variants, including six patients with syndromic hearing loss and six patients with known GJB2 variants. Novel pathogenic variants were noted in CHD7, ADGRV1, and ARID1B, with variants in the latter two genes identified in Hispanic patients. Overall, carriage of genetic variants was associated with better pre-CI audiometric thresholds at 2000 Hz (p = 0.048). On the other hand, post-CI thresholds were significantly worse in patients with inner ear malformations, particularly in patients with atretic cochlear nerve canals. CONCLUSION: Four novel pathogenic variants were identified, which contributes to knowledge of allelic spectrum for hearing loss especially in Hispanic patients. In this cohort, carriage of pathogenic variants particularly of GJB2 variants was associated with better pre-CI audiometric thresholds, while patients with inner ear malformations had worse post-CI audiometric thresholds.

The influence of perceptual and semantic categorization on inhibitory processing as measured by the N2-P3 response.

In daily activities, humans must attend and respond to a range of important items and inhibit and not respond to unimportant distractions. Our current understanding of these processes is largely based on perceptually simple stimuli. This study investigates the interaction of conceptual-semantic categorization and inhibitory processing using Event Related Potentials (ERPs). Participants completed three Go-NoGo tasks that increased systematically in the degree of conceptual-semantic information necessary to respond correctly (from single items to categories of objects and animals). Findings indicate that the N2 response reflects inhibitory processing but does not change significantly with task difficulty. The P3 NoGo amplitude, on the other hand, is attenuated by task difficulty. Further, the latency of the peak of the P3 NoGo response elicited by the most difficult task is significantly later than are the peaks detected during performance of the other two tasks. Thus, the level of complexity of conceptual-semantic representations influences inhibitory processing in a systematic way. This inhibition paradigm may be a key for investigating inhibitory dysfunction in patient populations.

3D for the people: multi-camera motion capture in the field with consumer-grade cameras and open source software.

Ecological, behavioral and biomechanical studies often need to quantify animal movement and behavior in three dimensions. In laboratory studies, a common tool to accomplish these measurements is the use of multiple, calibrated high-speed cameras. Until very recently, the complexity, weight and cost of such cameras have made their deployment in field situations risky; furthermore, such cameras are not affordable to many researchers. Here, we show how inexpensive, consumer-grade cameras can adequately accomplish these measurements both within the laboratory and in the field. Combined with our methods and open source software, the availability of inexpensive, portable and rugged cameras will open up new areas of biological study by providing precise 3D tracking and quantification of animal and human movement to researchers in a wide variety of field and laboratory contexts.